Knowledge-based dialogue for dynamic systems

As dynamic systems have become larger and more complex the problems associated with providing good dialogue systems have increased and there have been cases which have highlighted the need for better dialogues. Current dialogues are thought by operators to be inflexible and unstructured, the more an interface is automated the more it often hides what is really happening in the dynamic system. Such characteristics can be traced to the present limitations of communication devices and the ways in which communication and system knowledge are stored and accessed. There have been a number of advances recently in knowledge-based systems, dialogue architectures and computer graphics which could be used to provide improved interfaces. For example, techniques for representing both shallow and deep knowledge have been proposed, dialogue design has been improved by separating the functionality between Presentation, Dialogue Control and Application Modelling, and the Presentation has been considerably improved by enhanced colour graphics systems. The paper briefly discusses current difficulties in the control of dynamic systems and reviews useful techniques in artificial intelligence. The importance of human-centred design is emphasized and the advantages resulting from separation of low-level and high-level dialogue using knowledge-based dialogue architectures is discussed. Approaches to dialogue design in computer systems and dynamic systems are compared. Some current research progress and implementations in dynamic systems are reviewed.